mm: build_zonelists(): move clear node_load[] to __build_all_zonelists()
authorBo Liu <bo-liu@hotmail.com>
Tue, 18 Aug 2009 21:11:19 +0000 (14:11 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 18 Aug 2009 23:31:13 +0000 (16:31 -0700)
commit7f9cfb31030737a7fc9a1cbca3fd01bec184c849
treea8db215579bc073fb2b1d6d9690dbb78c79aa52d
parent503f7944fac68f4fdf71f8ebd06907f51eb64515
mm: build_zonelists(): move clear node_load[] to __build_all_zonelists()

If node_load[] is cleared everytime build_zonelists() is
called,node_load[] will have no help to find the next node that should
appear in the given node's fallback list.

Because of the bug, zonelist's node_order is not calculated as expected.
This bug affects on big machine, which has asynmetric node distance.

[synmetric NUMA's node distance]
     0    1    2
0   10   12   12
1   12   10   12
2   12   12   10

[asynmetric NUMA's node distance]
     0    1    2
0   10   12   20
1   12   10   14
2   20   14   10

This (my bug) is very old but no one has reported this for a long time.
Maybe because the number of asynmetric NUMA is very small and they use
cpuset for customizing node memory allocation fallback.

[akpm@linux-foundation.org: fix CONFIG_NUMA=n build]
Signed-off-by: Bo Liu <bo-liu@hotmail.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/page_alloc.c